A postnatal switch of CELF and MBNL reprograms in the developing heart

Auinash Kalsotraa, Xinshu Xiaob, Amanda J. Warda,c, John C. Castled, Jason M. Johnsond, Christopher B. Burgeb, and Thomas A. Coopera,c,e,1

Departments of aPathology, cMolecular and Cellular Biology, and eDevelopmental Biology, Baylor College of Medicine, Houston, TX 77030; bDepartment of Biology, Massachusetts Institute of Technology, Cambridge, MA 02139; and dRosetta Inpharmatics LLC, Merck and Company, Inc., 401 Terry Avenue North, Seattle, WA 98109

Edited by Eric N. Olson, University of Texas Southwestern Medical Center, Dallas, TX, and approved October 27, 2008 (received for review September 16, 2008) From a large-scale screen using splicing microarrays and RT-PCR, we birth and converts to adult function. This postnatal remodeling is identified 63 alternative splicing (AS) events that are coordinated in accomplished through transcriptional and posttranscriptional net- 3 distinct temporal patterns during mouse heart development. More works, including AS (16, 18). than half of these splicing transitions are evolutionarily conserved In the present study, splicing microarrays and computational between mouse and chicken. Computational analysis of the introns screens were used to investigate regulatory networks of AS in flanking these splicing events identified enriched and conserved vertebrate heart development. A large number of coordinated motifs including binding sites for CUGBP and ETR-3-like factors (CELF), splicing transitions were identified that undergo dramatic muscleblind-like (MBNL) and Fox proteins. We show that CELF pro- changes during heart development. Strikingly, Ͼ60% of the teins are down-regulated >10-fold during heart development, and splicing transitions tested were conserved between mouse and MBNL1 is concomitantly up-regulated nearly 4-fold. Using chicken, supporting functional relevance to heart development. transgenic and knockout mice, we show that reproducing the em- Computational and expression analyses, coupled with studies bryonic expression patterns for CUGBP1 and MBNL1 in adult heart using transgenic and knockout mice identified cis elements and induces the embryonic splicing patterns for more than half of the associated trans-acting factors that facilitate fetal-to-adult splic- developmentally regulated AS transitions. These findings indicate ing transitions including those regulated by postnatal changes in that CELF and MBNL proteins are determinative for a large subset of CELF and MBNL protein expression. Together, these analyses splicing transitions that occur during postnatal heart development. identify and characterize a highly conserved and highly regulated program of AS that supports postnatal growth and maturation of CUGBP and ETR-3-like factors ͉ heart development ͉ the developing heart. muscleblind-like ͉ splicing microarray Results oordinated control of alternative splicing (AS) on a genome- Global Analysis of AS During Mouse Heart Development. To identify Cwide scale has the potential to drive proteome transitions with AS transitions during mouse heart development, we carried out a wide-ranging and critical biological consequences (1, 2). Disruption large-scale screen using alternative splice-event profiling microar- of splicing and its regulation, therefore, is implicated in disease rays (described in ref. 15) and literature searches followed by causation and susceptibility (3). Splicing is regulated by RNA- RT-PCR validation. Splicing microarrays were used to detect changes in splicing and mRNA steady-state levels of 10,111 muscle- binding proteins that bind to cis-regulatory elements near the splice BIOLOGY and heart-enriched between embryonic day 17 (E17) and

sites. Some of the best-characterized splicing regulators include the DEVELOPMENTAL adult mouse heart RNA (supporting information (SI) Fig. S1A). serine–arginine (SR)-rich family, hnRNP proteins, and the Nova, Predicted AS changes were validated by RT-PCR (Fig. S1B). We PTB, FOX, TIA, CUGBP and ETR-3-like factors (CELF), and validated 147 splicing events that showed a Ͼ 2-fold change muscleblind-like (MBNL) families (4, 5). CELF and MBNL pro- between E17 and adult heart by microarray analysis. We focused on teins were first characterized as factors involved in the pathogenesis 54 events from microarray analysis and 9 events from the literature of myotonic dystrophy and were subsequently shown to be direct validated by RT-PCR as exhibiting Ն20-point change in percent regulators of AS (6–8). Recent advances in microarray and com- inclusion of the variably spliced region between E14 and adult putational technologies have allowed comprehensive analyses of heart. Among the 63 events collected, 41 (65%) exhibited an individual exons on a genome-wide scale, providing the ability to increase, and 22 (35%) exhibited a decrease in inclusion of the identify commonly regulated splicing events (9–12). variable region. Most variable regions (81%) were in-frame (mul- With some exceptions (13, 14), most large-scale analyses of tiples of 3 nt). The breakdown of different splicing modes (e.g., regulated splicing have focused primarily on differences between cassette exon, alternative 3Ј and 5Ј splice site, etc.) is provided in adult tissues and tissues/cell cultures depleted for a splicing regu- Table S1. lator rather than normal physiological transitions within a single To examine the relationship between transitions in splicing and tissue (9–11, 15). Developmental transitions provide an excellent opportunity to identify and determine the roles for coordinated splicing regulation associated with normal physiological change. Author contributions: A.K., J.M.J., C.B.B., and T.A.C. designed research; A.K., X.X., A.J.W., The vertebrate heart is particularly attractive for such analysis and J.C.C. performed research; A.K., X.X., J.C.C., J.M.J., C.B.B., and T.A.C. analyzed data; and because it undergoes extensive remodeling to meet the demands of A.K., J.C.C., C.B.B., and T.A.C. wrote the paper. increased workload in the developing organism (16). In addition, The authors declare no conflict of interest. the heart has relatively low cellular complexity and little cell This article is a PNAS Direct Submission. turnover (17) so that developmental splicing transitions reflect 1To whom correspondence should be addressed. E-mail: [email protected]. changes occurring within individual cells to a greater extent than in This article contains supporting information online at www.pnas.org/cgi/content/full/ many other tissues. The physiological changes that occur before and 0809045105/DCSupplemental. after birth are particularly important as the fetal heart adapts to © 2008 by The National Academy of Sciences of the USA

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0809045105 PNAS ͉ December 23, 2008 ͉ vol. 105 ͉ no. 51 ͉ 20333–20338 Downloaded by guest on September 26, 2021 Fig. 1. Subsets of AS transitions are coregulated during specific times of mouse heart development. Total RNA was isolated from 6–20 pooled hearts at the indicated time points. RT-PCR analysis was carried out for 63 AS events. Data are expressed as the percentage inclusion (Upper) and as the percentage of total change (Lower) for variable regions that show increased (A) or decreased (B) inclusion during development. Alternative exons are numbered according to Ensembl, and sequences of variable regions are presented in Table S2. In at least 2 independent assays for all E14 and adult samples, the standard deviation was Ͻ5 percentage points.

mRNA levels during heart development, we compared all validated Interestingly, 16 AS events exhibited biphasic transitions in which AS events that exhibited either a Ն20-point change in splicing Ͼ20% of the total change took place between E14 and E18, Ͻ20% and/or a Ն2.5-fold change in mRNA levels (78 genes total). Linear occurred between E18 and PN1, and Ͼ20% occurred between PN1 regression analysis revealed no significant correlation (R2 ϭ 0.19) and adult. The complete list of AS events assayed is provided in between the 2 datasets, indicating that different sets of genes are Table S2. regulated by AS transitions or mRNA levels (Fig. S1C). Compar- ative ontology analysis showed that different but overlapping A Large Fraction of Splicing Transitions Are Conserved During Mouse biological processes are associated with genes that undergo changes and Chicken Heart Development. To determine the level of conser- in splicing and those that undergo changes in mRNA levels (Fig. vation, AS was assayed during chicken heart development by using S2). Genes that undergo developmental splicing changes were orthologues of genes that exhibited changes in mouse and splicing enriched for cell structure and motility functions, but genes exhib- events from the literature. A total of 114 AS events were tested, and iting changes in mRNA levels were also enriched for signal trans- 51 of them exhibited Ն20-point change between E8 and adult duction and oxidative (lipid/steroid) metabolism. These results chicken heart (Table S3). Next, we performed a developmental demonstrate that during heart development, distinct subsets of time course of AS in chicken heart using E8, E12, E16, E20, post genes are regulated by changes in splicing and by changes in hatch day 2 (PH2), PH7, and adult hearts (Ͼ6 months). Just as most transcript levels. Similar results have been demonstrated in com- splicing transitions in mouse could be grouped according to max- parisons of splicing and mRNA profiles in adult tissues and during imal change relative to birth, most transitions during chicken heart T cell activation (11, 13, 15, 19). development could be grouped by maximal change relative to hatching (Fig. S3). Direct comparison of 48 splicing transitions of Subsets of Splicing Transitions Are Coordinated at Specific Times variable mRNA regions that are conserved between mouse and During Mouse Heart Development. To determine whether splicing chicken (Ն60% nucleotide identity) revealed that 32 (66.7%) were transitions are coordinated during mouse heart development, all 63 alternatively spliced and that 30 (62.5%) of these were similarly splicing events were assayed by RT-PCR using E14, E18, postnatal regulated in that the AS patterns change in the same direction day 1 (PN1), PN3, and adult (3 months) heart RNA. These results during development (Fig. 2A and Table S4). Furthermore, 21 of the were plotted both as percentage inclusion of the variable region(s) 30 AS transitions switched at comparable times during mouse and (Fig. 1 A and B Upper) to show the absolute change and as the chicken heart development (representative examples shown in Fig. percentage of total change (Fig. 1 A and B Lower) to identify the S4). Differences include 16 splicing events regulated only in 1 timing of maximal change. This analysis identified groups of splicing species and 2 transitions regulated in both species but in opposite transitions that are temporally coregulated. Splicing transitions directions. were grouped as early or late depending on whether the maximal analysis of the 30 conserved AS transitions change occurred before or after birth, respectively. For example, 18 indicated that most genes participate in processes such as develop- AS events exhibited a maximal increase in inclusion before birth ment, cell structure/motility, and muscle contraction (Fig. 2B). (Fig. 1A) whereas 12 exhibited an early decrease (Fig. 1B). Simi- Genes in the development group were further analyzed for their larly, of 17 postnatal splicing transitions, 10 exhibited increased annotated molecular function, showing that the majority of them inclusion, and 7 exhibited decreased inclusion of the variable region. encode cytoskeletal, signaling, or nucleic acid-binding proteins (Fig.

20334 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0809045105 Kalsotra et al. Downloaded by guest on September 26, 2021 Fig. 2. Most of the splicing transitions regulated during mouse heart develop- Fig. 3. Motifs significantly enriched (Upper) and/or conserved (P Ͻ 0.001) ment undergo similar transitions during chicken heart development. (A) Conser- among 8 mammalian species (Lower) in the 4 flanking 250-nt intronic regions of vation of AS in mouse and chicken heart development. (B) Gene ontology of 30 developmentally regulated mouse exons. Motifs that are significantly conserved events conserved between chicken and mouse. Enriched biological processes are and/or enriched that also exhibit a significant (P Յ 0.05) association with specific plotted on a Ϫlog (P value) scale with the threshold set to 1.3 [log (0.05)]. (C) temporal transitions in the regression analysis are indicated in red (decreasing Molecular functions assigned to the genes falling in the most significantly en- inclusion) or green (increasing inclusion) with icons indicating the transition riched process ‘‘Development’’ (P Ͻ 0.000005). pattern. Motifs in black are significantly conserved or enriched but not significant the regression analysis. ‘‘C’’ indicates that the motif is also significantly enriched among the developmentally regulated chicken exons. The 5 most significant 2C). Because the majority of the genes in this dataset do not pentamers are shown (see Table S5, Table S6, Table S7, and Table S8 for the undergo significant changes in mRNA levels or the reading frame, complete lists of motifs). Motifs resembling known binding sites of splicing factors (Table S10) are annotated. Fox (GCAUG) and CELF (GUGUG) motifs were these regulated events represent a conserved qualitative rather than significant in all analyses being enriched in mouse and chicken, conserved in quantitative change in the proteome. mammals, and associated with a specific temporal pattern.

Binding Motifs for Known Splicing Factors Including CELF and MBNL Are Enriched and/or Conserved in the Flanking Introns of Develop- the downstream intron. This pattern suggests that, like some other mentally Regulated Exons. Sequence motif analyses were conducted families of factors, the CELF family splicing factors may function to to identify putative regulatory elements in the flanking intronic activate or repress splicing in a location-dependent manner (20, 21). regions of developmentally regulated exons in mouse and chicken. The MBNL motif UGCGC was highly conserved in the last 250 The first and last 250 bases of the upstream and downstream introns bases of the upstream introns and was among the most significant were used in these analyses with the exclusion of the core splice-site motifs in the regression analysis being associated with decreased (ss) motifs (8 bases for 5Јss, 30 bases for 3Јss). Enrichment and exon inclusion compared with both E14 and adult data at all 5 time evolutionary conservation of pentanucleotide motifs were analyzed points. PTB motifs were identified more often in the enrichment

in the 4 regions comparing to first-order Markov models that analysis (in both mouse and chicken) than in the mammalian BIOLOGY account for dinucleotide effects on abundance and conservation conservation analysis, suggesting that PTB sites may evolve more DEVELOPMENTAL (see SI Text). To identify motifs associated with splicing transitions quickly. The regression analysis identified an association between occurring at specific times during heart development, a regression decreased exon inclusion with the ACUCC motif located in the first analysis between motif counts and AS changes at each time point 250 bases of the upstream intron. Finally, CUAAC and ACUAA was conducted for those pentamers that were significantly con- were highly conserved in multiple regions among mammals, and served or enriched. The most significant pentamers in the above CUAAC in the first 250 bases of the upstream intron was associated analyses are displayed in Fig. 3 (complete lists of significant motifs with decreased exon inclusion. These motifs resemble the motif are included in Table S5, Table S6, Table S7, Table S8, Table S9, (A)CUAAC that was found to be associated with skeletal and Table S10). muscle-specific exons (12, 22), and binding motifs for STAR A number of motifs in each intronic region were identified family proteins (23). including some that resemble the binding sites of known splicing factors. Two motifs in particular, GCAUG (Fox) and GUGUG A Postnatal Expression of CELF, MBNL, and Fox Proteins Correlates (CELF), located in the immediate downstream intronic region, with a Subset of AS Transitions. We found a significant enrichment/ were significant in all parameters examined: conserved among 8 conservation of binding motifs for the CELF, MBNL, and Fox mammalian species, enriched within this region in both mouse and families flanking the developmentally regulated exons. The two chicken, and a significant association with specific temporal changes CELF proteins expressed in heart, CUGBP1 and CUGBP2, are (increased exon inclusion late in development). The regression expressed at low levels in adult compared with embryonic heart analysis of additional CELF and Fox motifs enriched or conserved, (24). To determine the timing of CUGBP1 and CUGBP2 down- respectively, in this region also supports a role for these 2 proteins regulation in mouse heart development, we analyzed their protein in regulating developmental AS changes, especially in the postnatal and mRNA expression during the first 2 weeks after birth. As shown stages. Expression of CELF and Fox proteins during postnatal in Fig. 4A, CUGBP1 and CUGBP2 protein levels begin to decrease development are consistent with these results (see below). Inter- by PN6 and PN10, respectively. In contrast, MBNL1 protein levels estingly, CELF motifs were associated with increased exon inclu- begin to increase at PN5, whereas no consistent change was evident sion when located in the first 250 bases of the downstream intron in PTB levels. Strikingly, Fox-1 protein was robustly up-regulated and decreased exon inclusion when located in the last 250 bases of postnatally after PN6, whereas Fox-2 decreased slightly in adult

Kalsotra et al. PNAS ͉ December 23, 2008 ͉ vol. 105 ͉ no. 51 ͉ 20335 Downloaded by guest on September 26, 2021 heart development. In addition, our results suggest selective loss of CUGBP1 from the nucleus during heart development.

CUGBP1 and MBNL1 Regulate Distinct as well as Overlapping Subsets of Postnatal AS Transitions. Next, we sought to determine whether postnatal splicing transitions coincided with the changes in CELF and MBNL protein expression observed during postnatal develop- ment. The timing of 10 postnatal AS transitions selected for detailed analysis (8 late and 2 biphasic) demonstrated a good correlation with changes in CELF and MBNL1 protein expression. For instance, 8 of 10 AS transitions exhibited a maximal change after PN6 (Fig. 4F), the time at which CUGBP1 protein starts to decline and MBNL1 levels begin to rise. To establish causal relationships between CELF and MBNL protein expression and specific sets of AS transitions, we examined AS changes in adult hearts from mice overexpressing CUGBP1 or deficient in MBNL1 expression. Transgenic mice expressing Flag- tagged human CUGBP1 in a heart-specific and tetracycline- inducible manner were obtained by generating tet-inducible CUGBP1 (TRECUGBP1) lines and crossing these with heart- specific rtTA lines -Myh6-rtTA (see SI Text). TRECUGBP1/Myh6- rtTA bitransgenic animals given doxycycline (designated TgCUGBP1) expressed CUGBP1 protein 5–8 times above the endogenous levels in adult heart (close to levels in PN1 heart) and, importantly, MBNL1 levels were not altered in these heart tissues (Fig. S6A). MBNL1 is required for a subset of postnatal AS changes that occur during skeletal muscle development (25). To investigate the contribution of MBNL1 in regulating AS transitions in the developing heart, we used the previously characterized Mbnl1⌬E3/ ⌬E3 knockout mice (26). These mice exhibited no change in CUGBP1 protein levels in heart tissue from these mice (Fig. S6B). Fig. 4. Postnatal expression of CELF, Fox, and MBNL proteins correlate with AS We compared 44 developmentally regulated AS changes in adult ϩ Ϫ ⌬E3/⌬E3 transitions. (A) Steady-state protein levels. (B) Steady-state mRNA levels by RT- TgCUGBP1 mice ( and dox), Mbnl1 mice and their PCR. (C and D) Steady-state mRNA levels determined by real time RT-PCR (Taq- respective wild-type littermate controls and identified 24 AS events Man). (E) Nuclear and cytoplasmic distributions of CUGBP1, CUGBP2, Fox-1, Fox-2, that were altered by either increased expression of CUGBP1 and/or and MBNL1. The distribution of Fox-1 and 2 or GAPDH demonstrate the clean loss of MBNL1 (Fig. 5, all AS events screened are in Table S11). separation of nuclear and cytoplasmic fractions. (F) A tight postnatal time course Thirteen AS events exhibited a response only in CUGBP1 overex- demonstrates a correlation between the timing of postnatal AS transitions and pressing mice, 5 responded only in Mbnl1⌬E3/⌬E3 mice, and 6 changes in CELF, Fox, and MBNL expression. responded in both TgCUGBP1 and Mbnl1⌬E3/⌬E3 mice (Fig. 5 A–C and Fig. S7 A–C). All events reverted to the embryonic/early postnatal splicing pattern strongly suggesting that these events are mice when compared with PN1. Western blot analysis of serially primarily regulated by elevated CUGBP1 and/or decreased diluted PN1 protein samples compared with adult demonstrated MBNL1 activity in the embryo/early postnatal heart. Interestingly, Ϸ10-, 18-, and 3-fold reductions in CUGBP1, CUGBP2, and Fox-2 all splicing events regulated by both proteins exhibited antagonistic protein levels, respectively (Fig. S5). On the other hand, 10- and responses consistent with previous work (6, 24). Two AS events, 3.8-fold increases were observed in Fox-1 and MBNL1 proteins, Itga7 and VegfA, represent the 20 events that were unaffected by respectively. either CUGBP1 overexpression or Mbnl1 depletion (Fig. 5D and In contrast to the loss of CUGBP1 and CUGBP2 proteins, their Fig. S7D). Further supporting a determinative role for the postnatal mRNA levels remained unchanged during this developmental period expression of CUGBP1 and MBNL1 in regulating postnatal AS (Fig. 4 B and C). These results indicate that steady-state CUGBP1 and transitions, 18 of the 24 exons that responded to either CUGBP1 CUGBP2 protein levels are regulated posttranscriptionally, either at the and/or MBNL1 normally undergo either late or biphasic postnatal level of translation or protein stability (or both). Postnatal expression of transitions (Table S11). In addition, CELF- and MBNL-binding Fox-1, Fox-2, and MBNL1 mRNAs paralleled their respective protein motifs are significantly enriched in the upstream and downstream expression profiles (Fig. 4 B and D). introns flanking alternative exons regulated by CUGBP1 and MBNL1 has been shown to undergo a cytoplasmic to nuclear MBNL1, respectively (Fig. 5 E–H). These data indicate that over half (24 of 44) of splicing events regulated during heart develop- transition during the first 3 weeks of postnatal skeletal muscle ment are CUGBP1- and/or MBNL1-dependent, with the rest development (25). To determine whether CUGBP1, CUGBP2, or presumably affected by other splicing regulators. MBNL1 differed in nuclear-cytoplasmic distribution during post- natal heart development, we prepared nuclear and cytoplasmic Discussion fractions from PN2 and PN21 hearts. Loss of CUGBP1 protein at Conserved AS Transitions Are Part of the Remodeling Program During PN21 was seen only in the nuclear fraction, whereas the cytoplasmic Heart Development. Cascades of transcriptional changes are known levels remained unchanged (Fig. 4E). In contrast, CUGBP2 levels to coordinate regulatory networks during heart development (16). declined sharply in both nuclear and cytoplasmic fractions at PN21. Here, we present evidence that ensembles of variably spliced MBNL1 did not show nuclear accumulation as described in skeletal regions are also coordinately regulated during mouse and chicken muscle but rather exhibited a small increase in the cytoplasmic heart development. Particularly striking is the extent (Ͼ60%) to fraction at PN21. We conclude that MBNL1 is not primarily which the regulation of AS is conserved during heart development. controlled by changes in nuclear-cytoplasmic distribution during This is in contrast to large-scale comparisons of orthologous gene

20336 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0809045105 Kalsotra et al. Downloaded by guest on September 26, 2021 Fig. 5. A subset of postnatal splicing transitions are regulated by CUGBP1 and/or MBNL1. Forty-four AS events were tested in the hearts of TgCUGBP1 and Mbnl1⌬E3/⌬E3 mice and their respective littermate controls. (A–D) Splicing events were found to be regulated antagonistically by CUGBP1 and MBNL1 (A), regulated by CUGBP1 only (B), regulated by MBNL1 only (C), or CUGBP1 and MBNL1 independent (D). Each bar shows the mean Ϯ SD for the percent inclusion of the specified variable region. Statistical analysis was done by using 1-way ANOVA, followed by Tukey’s multiple-range test (P Ͻ 0.05). *, significantly different from E14. †, significantly different from wild-type littermates. (E–H) Computational analysis for enrichment of CELF- and MBNL-binding motifs in flanking introns of CUGBP1- and MBNL1-regulated exons. To the right of each plot are heat maps representing the P values for significant enrichment of indicated motifs in positions 12–250 of the upstream intron (upIn1), positions Ϫ250 to Ϫ31 of the upstream intron (upIn), positions 12 to 250 of the downstream intron (dnIn), and positions Ϫ250 to Ϫ31 of the downstream intron (dnIn1).

pairs in which Ͻ20% of cassette-type events are conserved between information between cells as the attachment points of costameric human and mouse (27–29). The unusually high level of conserva- filaments (31). tion observed here for both AS and its regulation most likely We also found that several proteins (10 of 36) that undergo reflects enrichment of functional splicing events, in contrast to developmentally regulated splicing transitions have nucleic acid-

genome wide surveys that are blind to functional consequences. binding properties. Six of these take part in RNA processing/ BIOLOGY

These results illustrate an advantage of performing analysis of metabolism, whereas 4 regulate transcription. Consistent with DEVELOPMENTAL splicing events associated with a specific physiological transition. identification of several splicing regulators among the set of devel- Consistent with studies demonstrating that conserved splicing opmentally regulated exons, we found examples of autoregulation events tend to maintain reading frame (28, 29), 25 of the 32 (78.1%) and cross-regulation of splicing regulators. MBNL1 autoregulates splicing events conserved between mouse and chicken maintain the splicing of its own exon 5, which encodes an 18-aa domain located reading frame. Taken together, these results strongly suggest that within an ultraconserved segment (32). This exon is included in the majority of these alternatively spliced isoforms have conserved fetal but not adult hearts of both mouse and chicken. Intriguingly, functions. The nonconserved splicing transitions are also of interest we find that the analogous exon in Mbnl2 (exon 8) is also within an as potential contributors to species-specific differences. ultraconserved region, exhibits conserved regulation in mouse and Gene Ontology (GO) analysis of conserved splicing transitions chicken, and is regulated by both CUGBP1 and MBNL1. CUGBP1 revealed particular enrichment for genes that are fundamental to also cross-regulates splicing of the splicing regulator Fox-2, which is the remodeling of cardiomyocytes during postmitotic growth, such highly expressed in brain, skeletal muscle, and heart (33). These as cytoskeletal rearrangement, nucleic acid binding and signaling. observations agree with and extend previous findings that auxiliary To supplement the GO analysis, we used literature searches to splicing regulators tend to cross-regulate (10, 25). identify functions for 36 of the 44 genes tested in TgCUGBP1 and Mbnl1⌬E3/⌬E3 mice (Fig. S8). Some of the proteins expressed from Postnatal Switch of CUGBP1 and MBNL1 Protein Expression Controls these genes have diverse cellular functions but many (15 of 36) Fetal-to-Adult Transitions for a Subset of Splicing Events. CELF- and associate with contractile apparatus either within the Z-disk or via MBNL-binding sites were found to be enriched/conserved among interactions with Z-disk structural or regulatory proteins. In both developmentally regulated AS events in mouse. Only 2 of the 6 avian and mammalian hearts, the force generating capacity rises CELF genes (CUGBP1 and CUGBP2) are expressed in heart and steeply during the first weeks after hatching or birth because of both proteins are down-regulated during heart development (24, accumulation and structural reorganization of contractile proteins 25). We demonstrated that CUGBP1 and CUGBP2 protein levels and a decrease in Ca2ϩ sensitivity (30). The Z-disk forms the drop by PN6 and PN10, respectively, and that regulation of both structural anchor for the sarcomere as the attachment point of the proteins is posttranscriptional. CUGBP1 protein stability is con- thin filaments and is crucial for communication of mechanosensory trolled by phosphorylation through a protein kinase C-dependent

Kalsotra et al. PNAS ͉ December 23, 2008 ͉ vol. 105 ͉ no. 51 ͉ 20337 Downloaded by guest on September 26, 2021 pathway, and a change in CUGBP1 phosphorylation during mouse Animals. Heart tissues were collected from wild-type FvB, Mbnl1⌬E3/⌬E3 (kindly heart development correlates with decreased protein stability in provided by Maurice Swanson, University of Florida College of Medicine, Gaines- adult tissue (34). Selective loss of nuclear CUGBP1 compared with ville, FL), transgenic TRECUGBP mice (design of mice is provided in SI Text) and CUGBP2 in 3-week-old postnatal hearts suggests that CUGBP2 chickens (White Leghorns) at indicated time points for subsequent analysis. might use a different mechanism of down-regulation. Our results Splicing Microarrays and RT-PCR Validations. Microarray studies were performed also indicate that MBNL1 expression is up-regulated during post- as described in SI Text. Microarray prediction data were used to design RT-PCR natal heart development via increased mRNA expression. primers around potential splicing regions. Percentage exon inclusion was calcu- Consistent with computational analyses showing enriched lated by using Kodak Gel logic 2200 and Molecular Imaging Software as: [(exon CELF- and MBNL-binding sites, we identified 24 developmentally inclusion band/(exon inclusion band ϩ exon exclusion band) ϫ 100]. regulated AS events that are sensitive to changes in the steady-state levels of CUGBP1 and/or MBNL1 in mouse heart. Thirteen of Computational Motif Analyses. Four intronic regions (Fig. 4) flanking devel- these events are CUGBP1-specific, 5 are MBNL1-specific, and 6 opmentally regulated exons in mouse and chicken were analyzed for motif conservation, enrichment, and correlation with time-course splicing alter- are antagonistically regulated by both CUGBP1 and MBNL1. All ation by using first-order Markov models as described in SI Text. 24 events revert to embryonic/early postnatal splicing patterns in ⌬E3/⌬E3 adult hearts of TgCUGBP1 and/or Mbnl1 mice. Eighteen of Western Blot Analysis and Real-Time RT-PCR. CUGBP1, CUGBP2, MBNL1, PTB, the 24 regulated exons undergo a postnatal switch, suggesting that Fox-1, Fox-2, and GAPDH protein expression by Western blot analysis and RNA they are normally regulated by CUGBP1 and/or MBNL1. Most expression by semiquantitative and real-time RT-PCR for was performed by TgCUGBP1 mice exhibit severe heart failure and arrhythmias, using standard procedures. causing high mortality rates. It remains to be determined which, if any, altered splicing event(s) phenocopy the cardiac defects seen in Statistics. Values are presented as mean Ϯ SD. Statistical significance was these mice. The cardiac phenotype of MBNL1⌬E3/⌬E3 mice remains determined by using 1-way ANOVA, followed by post hoc Tukey’s multiple range test (P Ͻ 0.05). to be characterized. Because misregulated expression of both CELF and MBNL is associated with misregulation of alternative splicing ACKNOWLEDGMENTS. We thank C. Armour for sample amplification and in myotonic dystrophy (35), it will be interesting to determine to microarray hybridization, D. Tran for initial validation, the what extent this postnatal splicing network is disrupted in the Sequencing Center (Baylor College of Medicine) for sequence confirmation of disease. RT-PCR products, Dr. R. Sorek (Weizmann Institute of Science, Rehovot, Israel) for information on AS in chicken, Dr. D. Black (University of California, Los ⌬E3/ Methods Angeles) for PTB antibodies, and Dr. M. Swanson for tissues from the Mbnl ⌬E3 line. This project was funded by the National Institutes of Health Grants Details of materials and methods used are in SI Text. R01GM076493 and R01HL45565 (to T.A.C.) and HG002439 (to C.B.B.).

1. Graveley BR (2001) Alternative splicing: Increasing diversity in the proteomic world. 20. Ule J, et al. (2006) An RNA map predicting Nova-dependent splicing regulation. Nature Trends Genet 17:100–107. 444:580–586. 2. Johnson JM, et al. (2003) Genome-wide survey of human alternative pre-mRNA splicing 21. Wang Z, Xiao X, Van Nostrand E, Burge CB (2006) General and specific functions of with exon junction microarrays. Science 302:2141–2144. exonic splicing silencers in splicing control. Mol Cell 23:61–70. 3. Wang GS, Cooper TA (2007) Splicing in disease: Disruption of the splicing code and the 22. Das D, et al. (2007) A correlation with exon expression approach to identify cis- decoding machinery. Nat Rev Genet 8:749–761. regulatory elements for tissue-specific alternative splicing. Nucleic Acids Res 35:4845– 4. Gabut M, Chaudhry S, Blencowe BJ (2008) SnapShot: The splicing regulatory machinery 4857. Cell 133:192 e1. 23. Galareau A, Richard S (2005) Target RNA motif and target mRNAs of the Quaking STAR 5. Matlin AJ, Clark F, Smith CW (2005) Understanding alternative splicing: Towards a protein. Nat Struct Mol Biol 12:691–698. cellular code Nat Rev Mol Cell Biol 6:386–398. 24. Ladd AN, Stenberg MG, Swanson MS, Cooper TA (2005) Dynamic balance between 6. Ho TH, et al. (2004) Muscleblind proteins regulate alternative splicing. EMBO J activation and repression regulates pre-mRNA alternative splicing during heart devel- 23:3103–3112. opment. Dev Dyn 233:783–793. 7. Philips AV, Timchenko LT, Cooper TA (1998) Disruption of splicing regulated by a 25. Lin X, et al. (2006) Failure of MBNL1-dependent post-natal splicing transitions in CUG-binding protein in myotonic dystrophy. Science 280:737–741. myotonic dystrophy. Hum Mol Genet 15:2087–2097. 8. Timchenko LT, et al. (1996) Identification of a (CUG)n triplet repeat RNA-binding 26. Kanadia RN, et al. (2003) A muscleblind knockout model for myotonic dystrophy. protein and its expression in myotonic dystrophy. Nucleic Acids Res 24:4407–4414. Science 302:1978–1980. 9. Blencowe BJ (2006) Alternative splicing: New insights from global analyses. Cell 27. Modrek B, Lee CJ (2003) Alternative splicing in the human, mouse and rat genomes is 126:37–47. associated with an increased frequency of exon creation and/or loss. Nat Genet 10. Boutz PL, Chawla G, Stoilov P, Black DL (2007) MicroRNAs regulate the expression of the 34:177–180. alternative splicing factor nPTB during muscle development. Genes Dev 21:71–84. 28. Sorek R, Shamir R, Ast G (2004) How prevalent is functional alternative splicing in the 11. Fagnani M, et al. (2007) Functional coordination of alternative splicing in the mam- human genome? Trends Genet 20:68–71. malian central nervous system. Genome Biol 8:R108. 29. Yeo GW, Van Nostrand E, Holste D, Poggio T, Burge CB (2005) Identification and 12. Sugnet CW, et al. (2006) Unusual intron conservation near tissue-regulated exons analysis of alternative splicing events conserved in human and mouse. Proc Natl Acad found by splicing microarrays. PLoS Comput Biol 2:e4. Sci USA 102:2850–2855. 13. Ip JY, et al. (2007) Global analysis of alternative splicing during T-cell activation. RNA 13:563–572. 30. Siedner S, et al. (2003) Developmental changes in contractility and sarcomeric proteins 14. Boutz PL, et al. (2007) A post-transcriptional regulatory switch in polypyrimidine from the early embryonic to the adult stage in the mouse heart. J Physiol 548:493–505. tract-binding proteins reprograms alternative splicing in developing neurons. Genes 31. Samarel AM (2005) Costameres, focal adhesions, and cardiomyocyte mechanotrans- Dev 21:1636–1652. duction. Am J Physiol 289:H2291–H2301. 15. Castle J, et al. Differential expression of 24,426 human alternative splicing events and 32. Bejerano G, et al. (2004) Ultraconserved elements in the human genome. Science predicted cis-regulation in 48 tissues and cell lines Nat Genet, doi:10.1038/ng64. 304:1321–1325. 16. Olson EN (2006) Gene regulatory networks in the evolution and development of the 33. Underwood JG, Boutz PL, Dougherty JD, Stoilov P, Black DL (2005) Homologues of the heart. Science 313:1922–1927. Caenorhabditis elegans Fox-1 protein are neuronal splicing regulators in mammals. 17. Pasumarthi KB, Field LJ (2002) Cardiomyocyte cell cycle regulation. Circ Res 90:1044– Mol Cell Biol 25:10005–10016. 1054. 34. Kuyumcu-Martinez NM, Wang GS, Cooper TA (2007) Increased steady-state levels of 18. Xu X, et al. (2005) ASF/SF2-regulated CaMKIIdelta alternative splicing temporally CUGBP1 in myotonic dystrophy 1 are due to PKC-mediated hyperphosphorylation. Mol reprograms excitation–contraction coupling in cardiac muscle. Cell 120:59–72. Cell 28:68–78. 19. Pan Q, et al. (2004) Revealing global regulatory features of mammalian alternative 35. Ranum LP, Cooper TA (2006) RNA-mediated neuromuscular disorders. Annu Rev splicing using a quantitative microarray platform. Mol Cell 16:929–941. Neurosci 29:259–277.

20338 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0809045105 Kalsotra et al. Downloaded by guest on September 26, 2021